MicroRNA (miRNA), one of non-coding RNAs (ncRNAs), regulates gene expression directly by arresting the messenger RNA (mRNA) translation, which is important for identifying putative miRNAs. In this study, we suggest a searching procedure for human miRNA precursors using genetic programming that automatically learn common structures of miRNAs from a set of known miRNA precursors. Our method consists of three-steps. At first, for each miRNA precursor, we adopted genetic programming techniques to optimize the RNA Common-Structural Grammar (RCSG) of populations until certain fitness is achieved. In this step, the specificity and the sensitivity of a RCSG for the training data set were used as the fitness criteria. Next, for each optimized RCSG, we collected candidates of matching miRNA precursors with the corresponding grammar from genome databases. Finally, we selected miRNA precursors over a threshold (=365) of scoring model from the candidates. This step would reduce false positives in the candidates. To validate the effectiveness of our miRNA method, we evaluated the learned RCSG and the scoring model with test data. Here, we obtained satisfactory results, with high specificity (= 51/64) and proper sensitivity (= 51/82) using human miRNA precursors as a test data set. © Springer-Verlag Berlin Heidelberg 2004.
CITATION STYLE
Nam, J. W., Lee, W. J., & Zhang, B. T. (2004). Computational methods for identification of human microRNA precursors. In Lecture Notes in Artificial Intelligence (Subseries of Lecture Notes in Computer Science) (Vol. 3157, pp. 732–741). Springer Verlag. https://doi.org/10.1007/978-3-540-28633-2_77
Mendeley helps you to discover research relevant for your work.